期刊
INFORMATION SCIENCES
卷 550, 期 -, 页码 71-90出版社
ELSEVIER SCIENCE INC
DOI: 10.1016/j.ins.2020.10.043
关键词
Three-way decision; Dynamic three-way approximations; Incremental; Fuzzy probability
资金
- National Natural Science Foundation of China [61773324, 61876157, 61976182]
- Humanity and Social Science Youth Foundation of Ministry of Education of China [20YJC630191]
- Fundamental Research Funds for the Central Universities [JBK2001004]
- Fintech Innovation Center of Southwestern University of Finance and Economics
- Financial Intelligence & Financial Engineering Key Laboratory of Sichuan Province
This paper focuses on the integration of dynamics and fuzziness in the context of three-way approximations in the fuzzy probability space. By analyzing fuzzy conditional probability functions and designing incremental algorithms, a dynamic three-way approximation model based on the principle of least cost is proposed.
As a special model of three-way decision, three-way approximations in the fuzzy probability space can be interpreted, represented, and implemented as dividing the universe into three pair-wise disjoint regions, i.e., the positive, negative and boundary regions, which are transformed from the fuzzy membership grades with respect to the fuzzy concept. To consider the temporality and uncertainty of data simultaneously, this paper focuses on the integration of dynamics and fuzziness in the context of three-way approximations. We analyze and investigate three types of fuzzy conditional probability functions based on the fuzzy T-norm operators. Besides, we introduce the matrix-based fuzzy probability decision-theoretic models to dynamic three-way approximations based on the principle of least cost. Subsequently, to solve the time-consuming computational problem, we design the incremental algorithms by the updating strategies of matrices when the attributes evolve over time. Finally, a series of comparative experiments is reported to demonstrate and verify the performance of proposed models. (C) 2020 Elsevier Inc. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据